Prediction of Two-Element Cylindrical Electrostatic Lens Parameters Using Dynamic Artificial Neural Network

Abstract

Two-element cylindrical electrostatic lens systems allowed to control low energy electron or charged particle beam have great importance. In this context, dynamic articial neural network using nonlinear autoregressive exogenous model has been utilized to predict optimum linear magnication and overall voltage values for these lens designs. The focusing characteristics of electron beam in two-element cylindrical lens systems are investigated with two dierent nonlinear autoregressive exogenous based articial neural network models. First articial neural network model is employed for predicting of voltage ratios of lenses and magnication values. This model interpolates among the object (P) and image positions (Q) and nally nds optimum voltage ratios and magnication values through training dataset. Due to the deviations of electron trajectories in a real lens system, the spherical aberration eects are also taken into account to determine the optimal lens parameters. Therefore, the second articial neural network model is constructed for predicting spherical aberration coecients in image point. For each of articial neural network models, training, test and validation data set are obtained from SIMION 8.1 ion and electron optics software. Articial neural network model outputs are compared with the SIMION data and very good agreements are found. While articial neural network is frequently applied in dierent elds, this is the rst study that uses dynamic articial neural network to predict the parameters of electrostatic lens. It is believed that this pioneering work will be a guide for the future investigations in lens design systems.

Analysis of Radial Dependence of the Localized Magnetic Field using Artificial Neural Networks

“…In recent years, artificial neural networks have seen an explosion of interest in atomic and molecular physics [12][13][14][15][16]. Artificial neural networks (ANNs) are parallel computing systems which are modelled on the behavior of neurons in the human brain.…”

Prediction of First Order Focusing Properties of Ideal Hemispherical Deflector Analyzer Using Artificial Neural Network

“…In the absence of any fringing fields, the ideal HDAs indicate the first order focusing characteristics. In recent years, successful applications are performed for solving problems in charged particle optics using artificial neural networks (ANNs) in prediction [14,15] and classification [16][17][18]. ANNs are parallel computing systems with interconnected processors.…”

Classification of Electron Gun Operation Modes Using Artificial Neural Networks

“…It is a remarkable fact that recently artificial intelligence technique has achieved significant advantages in optimal design of electrostatic lenses [7][8][9][10][11]. Artificial intelligence is a mathematical invention that simulates human intelligence [12].…”